Keras optimizer


an optimizer and a loss can be defined in the node "DL Python network creator" with model.compile(...), and in the node "DL Keras Network Learner". Does the second supersedes the first one?



Hi Francesco,

Yes, this is currently the case. However, we plan to respect precompiled models in the learner node in a future version of our Keras integration. (E.g. the model would only be recompiled if the user chose a different optimizer/loss in the learner dialog or the like.)


This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.